This project will define the role of the basal ganglia in selective attention, and is motivated by a consideration of an emerging understanding of fronto-striatal circuits. Although Parkinson patients have motor problems, consideration of neuroanatomy suggests a more general description of the computation formed by the basal ganglia. One neural circuit begins in dorsolateral prefrontal cortex, an area often thought to subserve working memory for the separate domains of color, form, and space. The circuit courses through the basal ganglia in a manner not unlike that for circuits coming from motoric frontal regions, suggesting that the basal ganglia performs the same computation for each circuit. We propose an attentional computation in which the basal ganglia are involved in establishing, maintaining, and switching set. A decisive test of the theory would come from a demonstration that Parkinson patients and frontal stroke patients have impairment in establishing and switching sets involving color and form. Previous studies of Parkinson patients in our laboratories and others have not provided evidence for a role of the basal ganglia in shifting spatial attention, or shifting between movement sequences. We therefore draw a distinction between biasing functions of attention and selective functions, and suggest that the failure of earlier studies to identify attentional functions for basal ganglia was due to their use of biasing paradigms. These paradigms cue for one stimulus feature or dimension, leaving others as still relevant should they unexpectedly occur. Selection, on the other hand, requires that irrelevant dimensions be actively blocked to prevent their activation of a response in a given task context. Since the basal ganglia contain both inhibitory and net excitatory subcircuits, this dual characteristic fits with our proposal that the basal ganglia involves both selection of relevant dimensions and inhibition of irrelevant ones. The theory of basal ganglia attention switching within fronto-striatal circuits will be tested in Parkinson patients by manipulations of medication state, and in patients with infarcts of dorsolateral prefrontal cortex in a series of experiments on attention to color and form.